Raymond Toy wrote:
> I've played around a bit with quadpack and have a very rough
> interface to quadpack for maxima. Quadpack is an adaptive
> numerical quadrature package.
[...]
> Any interest in this package? Should it be distributed
> separately from maxima? A part of maxima? If a part, where?
That sounds like a great idea to me. I have used Quadpack
for calculating integrals (marginal probability densities)
and I am happy with the results. It is especially useful
that Quadpack includes routines for integration over
unbounded intervals. By the way, how did you make the
interface -- translate Fortran to Lisp or call
the Fortran directly?
My recommendation is to make Quadpack an add-on library.
(If there isn't a convenient library interface, we should
make one.) Instead of making Maxima proper bigger and bigger,
I think additional functions can often be put into libraries.
This makes it easier for more people to contribute.
(Incidentally this is the development model of the R project
and, indeed, many people have contributed add-on libraries.)
I would also recommend that the existing numerical
integration routines be deprecated. Not that there's
anything wrong with them, but I think it's best to
avoid confusion; to be honest I don't usually believe
"there's more than one way to do it". Just my $0.02.
For what it's worth,
Robert Dodier
__________________________________
Do you Yahoo!?
Free Pop-Up Blocker - Get it now
http://companion.yahoo.com/